# Does "Lifting" Lifted Product Codes Reduce Fault-Tolerant Overhead?

A new arxiv preprint from Yuta Hirasaki and Jong Yeon Lee answers yes — and demonstrates that a systematic group-extension construction for lifted product (LP) codes can simultaneously improve code parameters *and* lower the space overhead of [fault-tolerant quantum computing](https://quantumintel.tech/glossary/fault-tolerant-quantum-computing) operations. The paper, posted July 30 2026, addresses one of the most practically consequential open problems in quantum error correction (QEC): how to scale qLDPC code families while keeping their Tanner graph structure local and their logical-gate gadgets tractable.

The core result has three parts. First, the authors report LP codes with better code parameters than any previously reported in the literature — though the paper does not specify exact parameter triples in the abstract, the claim is that their systematic construction outperforms ad-hoc prior constructions. Second, they show that code-surgery gadgets — the connective tissue that lets engineers perform [logical qubit](https://quantumintel.tech/glossary/logical-qubit) operations between code blocks — can be transferred across finite lifts through chain maps, in several cases at lower space overhead than baseline. Third, they propose "lifting" as a principled first step for constructing thermodynamic families of algebraically defined qLDPC codes that lack an underlying Euclidean lattice, a class that includes many of the most promising near-term fault-tolerant code candidates.

---

## What Are Lifted Product Codes and Why Do They Matter?

LP codes are a family of quantum low-density parity-check (qLDPC) codes constructed by taking a product of two classical LDPC codes defined over a group algebra. They sit at the intersection of algebraic coding theory and quantum error correction, and they matter for one blunt reason: they encode more [logical qubits](https://quantumintel.tech/glossary/logical-qubit) per physical qubit than surface codes, while maintaining bounded check weight — meaning each stabilizer measurement touches only a fixed, small number of physical qubits regardless of code distance.

The surface code, the current hardware favorite at [IBM Quantum](https://quantumintel.tech/companies/ibm) and [Google Quantum AI](https://quantumintel.tech/companies/google-quantum-ai), encodes exactly one logical qubit per code block and requires physical qubit counts that scale as O(d²) for distance d. Good qLDPC codes, including LP codes, can achieve encoding rates and distances that are asymptotically superior. The engineering catch has always been that implementing fault-tolerant logical operations — transversals, code surgery, lattice surgery analogues — is considerably harder when the code's geometry doesn't map cleanly onto a 2D grid.

Hirasaki and Lee's construction directly attacks that engineering catch.

---

## The Construction: Group Extensions and Graph Lifts

The technical mechanism is a two-step process. A "lift" of a graph is a covering space construction: you take a base graph and produce a larger graph that maps back onto it in a structured way, preserving local neighborhoods. The key insight of the paper is that choosing lifts via *group extensions* — a classical algebraic structure that interpolates between direct products and semidirect products of groups — gives you systematic control over how code parameters and logical operators transform across the family.

Concretely, the chain and cochain maps that relate codes within a lift family allow the authors to:

1. **Track logical operator evolution.** When you lift a code, you need to know what happens to the logical X and Z operators. The chain map framework makes this explicit and computable rather than requiring ad-hoc rederivation for each new code.

2. **Transfer surgery gadgets.** Code surgery — cutting, joining, and deforming code blocks to implement logical gates — normally has to be redesigned from scratch for each code instance. The authors show that for "selected finite lifts," surgery gadgets defined on a base code can be transported to larger codes in the family. In several cases, the transported gadget has *lower space overhead* than naively constructing it fresh.

3. **Build parallel product surgery.** For the specific subclass of lifted clustered cyclic codes, the authors develop a parallel product surgery protocol, which is relevant for reducing the time cost of logical gate sequences.

---

## Thermodynamic Families: A Less-Covered Contribution

The final and arguably most forward-looking contribution is the proposal to use lifting as the opening move in defining *thermodynamic families* for algebraically defined qLDPC codes. A thermodynamic family, in the sense used here, is a sequence of codes whose [error threshold](https://quantumintel.tech/glossary/error-threshold) behavior can be characterized by finite-size scaling — the same framework used to locate phase transitions in statistical mechanics.

This matters because, for surface codes defined on Euclidean lattices, thermodynamic families are well-understood and give practitioners a clear picture of how decoding performance evolves with code size. For qLDPC codes without a geometric lattice embedding — a category that includes some of the most parameter-efficient constructions known — no analogous systematic framework existed.

The authors test their proposal on several base codes and selected lifts, computing coherent information (a standard QEC diagnostic for logical channel quality). They observe finite-size crossings — signatures consistent with a phase transition and threshold behavior. Importantly, they are careful to note that their results also indicate additional conditions beyond lifting alone are needed to uniquely define such a family. That intellectual honesty is notable; the framework is promising but incomplete, and the paper says so explicitly.

---

## Skeptical Read: What's Still Missing

The abstract is light on specifics. The claim of "better code parameters than previously reported" is the most consequential in the paper, but without the actual [[n, k, d]] triples visible in the abstract, external verification requires reading the full paper. That's normal for a preprint, but it means the performance claims should be treated as preliminary until peer-reviewed or independently reproduced.

The surgery overhead reduction result is qualified — it applies in "several cases" for "selected finite lifts," not universally. Engineers evaluating this for hardware implementation will need to identify which specific lift families correspond to codes they can actually build on their target platform.

The thermodynamic family result is explicitly flagged as incomplete by the authors themselves. The finite-size crossings they observe are necessary but not sufficient to declare a unique thermodynamic family defined.

---

## Industry Trajectory

The broader qLDPC race is intensifying. [Microsoft Quantum](https://quantumintel.tech/companies/microsoft) has publicly committed to topological qubits with a QEC stack built around high-rate codes. [Quantinuum](https://quantumintel.tech/companies/quantinuum) has demonstrated logical qubit operations in trapped-ion hardware using QEC codes with non-trivial structure. Startups and academic groups are under competitive pressure to deliver codes that are not just theoretically superior on paper but operable with realistic decoders and gate gadgets on near-term hardware.

Hirasaki and Lee's work is a theorist's contribution to a problem that is rapidly becoming an engineer's bottleneck. The value isn't a single better code — it's a *systematic method* for producing families of better codes with known logical-gate structure. If the surgery overhead reductions hold up across a broader class of lifts, this could inform how hardware vendors design their QEC layers as physical qubit counts scale into the thousands and beyond.

---

## Key Takeaways

- Hirasaki and Lee (arXiv, July 30 2026) introduce a group-extension and graph-lift construction for lifted product (LP) qLDPC code families.
- The construction is claimed to yield LP codes with better code parameters than previously reported, with code relationships tracked via chain and cochain maps.
- Code-surgery gadgets can be transferred across finite lifts, with lower space overhead in several demonstrated cases — a direct engineering benefit for fault-tolerant logical operations.
- Parallel product surgery is developed specifically for lifted clustered cyclic codes.
- The paper proposes lifting as a systematic first step toward defining thermodynamic families for algebraically defined qLDPC codes without Euclidean lattice structure; finite-size coherent information crossings are observed, but the authors note additional conditions are required for uniqueness.
- Specifics on improved code parameter triples and the breadth of overhead reductions require reading the full paper; claims are preliminary pending peer review.

---

## Frequently Asked Questions

**What is a lifted product (LP) code?**
A lifted product code is a quantum error-correcting code constructed from a product of two classical LDPC codes defined over a group algebra. LP codes are a subclass of qLDPC codes notable for favorable ratios of logical to physical qubits and bounded stabilizer weight.

**How does "lifting" improve LP codes in this paper?**
Hirasaki and Lee use group extensions and graph-covering (lift) constructions to systematically generate larger LP codes from base codes. The construction preserves local Tanner graph structure while enabling transfer of logical operators and surgery gadgets via chain maps, in some cases at lower space overhead.

**What is code surgery and why does overhead reduction matter?**
Code surgery is a technique for implementing logical gate operations between qLDPC code blocks by temporarily merging or splitting codes. Reducing space overhead means fewer ancilla physical qubits are required during the operation — directly lowering hardware resource requirements for fault-tolerant computation.

**What is a thermodynamic family of QEC codes?**
A thermodynamic family is a sequence of codes whose error threshold behavior can be characterized by finite-size scaling, analogous to phase transitions in physics. Establishing such families for qLDPC codes would give practitioners predictive models for how decoding performance scales with code size.

**Is this work ready for hardware implementation?**
Not directly. The paper is a theoretical construction paper. Its value is in providing systematic methods and improved code candidates that hardware teams and decoder developers can evaluate. The surgery overhead improvements apply to selected cases, and the thermodynamic family framework is acknowledged as incomplete.